Advances and Challenges in Earnings Quality Research

1. Earnings quality research: Advances, challenges and future researchMark L . Defond Journal of Accounting and Economics 50 (2010) 402–409

2. Abstract This discussion makes several observations regarding the earnings quality research reviewed in Dechow, Ge and Schrand (2010) (DGS). I discuss some of the factors that led to the large growth in the earnings quality literature over the past two decades, and note a few of the important contributions from this literature. I also present what I view as several major challenges the literature faces as well as some avenues for future research. In addition, I discuss the difficulties in evaluating such a diverse body of literature, and comment on DGS’s major conclusions.

3. Growth in the EQ literature Research on EQ has grown dramatically over the past two decades. This is particularly true in the earnings management area, which comprises the majority of studies reviewed by DGS. While speculative, I believe the major drivers of the growth in the EQ area are likely to be the confluence of several factors that have both encouraged and facilitated this line of research. A factor encouraging this research was the SEC’s harsh allegations during the 1990s of widespread earnings management among public companies (e.g., Levitt, 1998). These assertions depicted managers as routinely engaging in opportunistic earnings management aimed at meeting capital market expectations. Growth in the EQ literature, and particularly earnings management research, continued to accelerate in thewake of the high profile accounting frauds in the early 2000s that culminated in the passage of the Sarbanes–Oxley Act of 2002 (SOX).

4. Growth in the EQ literature • The SEC allegations were also highly critical of the auditing profession, whom they portrayed as managers’ willing accomplices in deceiving the public. Such criticisms encouraged researchers to expand the scope of their studies to include the impact of auditors’ incentives on earnings management. This eventually led to a line of research investigating the role of audit quality on EQ issues beyond simply earnings management. • More recently, the development and implementation of a set of internationally accepted accounting standards has also stimulated growth in the EQ literature. The International Accounting Standard Board’s (IASB) explicit objective of developing a set of ‘‘high quality’’ accounting standards has naturally focused researchers’ attention on fundamental issues related to EQ. • In addition, the widespread adoption of International Financial Reporting Standards (IFRS) has generated interest in international research that compares accounting practices across countries.

5. Growth in the EQ literature Another factor that I believe helped fuel the growth in EQ research is the introduction of the abnormal accruals model in Jones (1991).While it has been and continues to be controversial, the Jones model is noteworthy for providing the literature with a more-or-less generally accepted measure of abnormal accruals. Finally, the introduction of a myriad of new electronic databases has facilitated growth in large sample EQ studies in areas that had previously been limited due to the high costs of hand collecting data (e.g., in the areas of corporate governance, executive compensation, auditing, and the international area, just to name a few).

6. Progression of and accomplishments in the EQ literature As it has expanded, the EQ literature has improved on several dimensions. One of the most dramatic improvements has been in the rigor of the EQ proxies, particularly the proxies measuring abnormal accruals. Abnormal accruals is by far the single most popular proxy used in the papers reviewed in DGS and a hallmark of the EQ research is the nearly continual innovations made to these proxies. Indeed, the measure used in the earliest study that employs abnormal accruals as a proxy (Healy, 1985), bears little resemblance to the proxies used in the current literature, which consist primarily of modified versions of the model in Jones (1991). EQ researchers are also continually developing new models that compete with the family of models descended from Jones (1991). Competition among models is healthy in terms of moving the literature forward, but most competitors do not survive. Relatively recent exceptions are two abnormal accruals proxies gaining acceptance in the literature: the proxy introduced in Dechow and Dichev (2002), and one of its variants, developed in Francis et al. (2005).

7. Progression of and accomplishments in the EQ literature While formulating policy is not the goal of the EQ literature, or of positive research in general, there is evidence that is consistent with EQ research influencing standard setters and lawmakers. For example, a recent report on audit quality by the US Department of the Treasury (2008) references numerous EQ studies (e.g., Ogneva et al., 2007; Myers et al., 2003). The Treasury Department also recently published a commissioned study by an academic researcher that summarizes the EQ literature on restatements (i.e., Scholz, 2008). In addition, the Congressional debates leading up to the passage of the Sarbanes–Oxley Act of 2002 cite several academic studies (e.g., DeFond et al., 2002). Finally, I am aware of several cases in which the Treasury Department and the FASB have sought informal input directly from accounting academics regarding research studies that potentially inform proposed standards (e.g., Dechow et al., 1996; Hanlon et al., 2008). Thus, there is ample evidence that policy makers are very much aware of the EQ literature and that it plays a role in the policy making process.

8. Challenges in the EQ literature • While I think that the literature has made important advances, I also think there are many challenges in studying EQ. One is that all of our abnormal accruals models suffer from the inherent limitation that we are unable to validate the accuracy of their predictions. • For example, we are unable to verify whether our estimates of discretionary accruals are the result of management’s opportunistic accounting choices, or just an artifact of the particular model we are using. This is a construct validity problem, which means that we are unsure whether these proxies really measure the underlying theoretical constructs they are intended to measure. • While creating proxies for phenomena that cannot be directly observed is common in the social sciences, it has important implications for EQ research. One implication is that every test using abnormal accruals proxies is a joint test of the hypothesis related to the research question and the hypothesis that the proxy is a valid measure. Another implication is that it presents difficulties in evaluating which proxy is ‘‘best’’, at least on the dimension of construct validity.

9. Challenges in the EQ literature • Earnings restatements and SEC Accounting and Auditing Enforcement Releases (AAERs) are potentially attractive alternatives to abnormal accruals as a proxy for EQ. One perceived advantage of restatements and AAERs over abnormal accruals is that they appear to be more direct proxies for EQ. • One such limitation is related to selection bias. In the case of restatements, they only capture poor EQ that has been both discovered and judged by the firm to merit restatement. In the case of AAERs, limited resources provide the SEC with incentives to pursue only the most egregious and high profile securities law violators (Pitt and Shapiro, 1990). • Taken together, these selection criteria mean that restatements and AAERs may not be appropriate for EQ studies that wish to capture unintentional errors or earnings management that either falls within the parameters of GAAP, or that goes undetected.

10. Future directions for EQ research Going forward, I think there are accounting developments on the horizon about which we know relatively little, and hence are logical prospects for future research. One example is fair-value accounting, which represents a potentially sea-changing development in the accounting environment. Fair-value accounting essentially follows from the explicit rejection of the concept of conservatism by standard setters in the newly proposed conceptual framework for financial reporting (IASB, 2008). Changing to a fair-value-based accounting system portends a marked shift from the traditional historical cost-based accounting model upon which our existing EQ research is based. How a fair value accounting model is likely to impact EQ, and what EQ will look like under such a model is very much an open question. Thus, this is an area where future research may be particularly beneficial. SOX also continues to provide opportunities for EQ research. While there are already a large number of SOX studies, many of them were done before SOX is likely to have reached a state of equilibrium.

11. Future directions for EQ research EQ research is also likely to benefit from further exploring the distinctive features of EQ in the debt markets, currently only a small area of the literature. While research on the role of accounting in debt contracting is certainly not new, there is relatively little research on the characteristics of EQ with respect to decision makers in debt markets. Both the growth in debt markets internationally and the recent debt crisis provide strong motivation for further research in this area. Another potential source of beneficial future research comes from areas that have been somewhat neglected in the literature. One of these areas is fraudulent financial reporting. A strong message communicated by the rapid and enthusiastic adoption of SOX is that users, or at least lawmakers, are especially averse to earnings management that crosses the line into fraudulent financial reporting.

12. Future directions for EQ research A related area for potential future research is gaining a better understanding of how the EQ literature impacts policy making. While there is ample evidence that policy makers are aware of the EQ literature (as discussed above), it is unclear whether or how EQ research actually influences policy makers’ decisions. Finally, an area of the EQ literature that seems relatively under-researched is so-called ‘‘real activities’’ manipulation, or ‘‘transaction’’ management. The fundamental importance of this area is evidenced in Graham et al. (2005), which concludes that earnings management is not only widely practiced, but that the majority of earnings management results from manipulating real operating activities. Transaction management through select activities such as manipulating R&D expenditures and asset sales has been investigated in several prior studies (e.g., Baber et al., 1991; Dechow and Sloan, 1991; Bartov, 1993).

13. SFAC #1 “Objectives of Financial Reporting by Business Enterprises” Statement of Financial Accounting Concepts – SFAC 1 The objectives of the conceptual framework is to provide information: • Useful for decision making. • That helps predict cash flows. • About economic resources, claims to resources, and changes in resources and claims

14. Comments on DGS • While the authors attempt to capture a commonality by using the definition of EQ provided in Statement of Financial Accounting Concepts (SFAC) No. 1, this definition is quite broad. As a result, the EQ studies evaluated in DGS really consist of a variety of heterogeneous research areas, which presents problems in trying to compare construct validity across studies. For example, one stream of the EQ literature is concerned with measuring ‘‘opportunistic earnings management,’’ while another stream is concerned with measuring earnings ‘‘predictability.’’ • Another difficulty in drawing conclusions about the EQ proxies is the wide variation in the way the proxies are measured. For example, while ‘‘abnormal accruals’’ is often used to proxy for the underlying theoretical construct of earnings management, it can be measured very differently across studies (e.g., Jones model discretionary accruals versus accruals from the Dechow and Dichev model, working capital versus total accruals, and signed versus absolute values). This means that the proxy labeled ‘‘abnormal accruals’’ essentially captures many different proxies.

16. AUDIT TRIANGULATION We focus on the condition in which two kinds of management-controlled audit evidence – evidence from the financial statements and evidence from internal data depicting performance of a key business process – is contradicted by external evidence suggesting that a key business objective has not been attained. Dem. Financeiros ??? Dados Internos Evidencias Externas

17. Comments on DGS • The authors are essentially implying that what has become popularly termed ‘‘triangulation’’ is sometimes taken too far. While the authors do not cite specific cases of ‘‘improper’’ triangulation in the literature, I believe that the findings in Carey and Simnett (2006) indirectly exemplify this notion. Carey and Simnett (2006) finds that auditor tenure is positively associated with the propensity to issue going concern opinions (consistent with increased auditor independence), but not with the level of discretionary accruals. • While one view of these findings is that the discretionary accruals results do not triangulate with the going concern results, Carey and Simnett (2006) argue that it is not clear that they should. The crux of their argument is that auditors have much greater control over their audit opinion decision than they do over management’s discretionary accruals choices. As long as the client is proposing within-GAAP discretionary accruals, the client is likely to prevail.

18. Summary • I also suggest that a major advancement in the EQ literature is the increased rigor and continual evolution of the abnormal accruals proxies. I further point out that the literature has generated new knowledge regarding a variety of factors that help explain EQ – such as earnings target incentives, country-level institutions, audit quality, internal controls, and timely loss recognition. • I also caution that the EQ literature faces several inherent challenges, including construct validity problems in the abnormal accruals proxies and difficulties in drawing inferences from EQ studies about decision-maker preferences. • In addition, I suggest several areas for future research. These include the effects on EQ of the shift toward a fair value based accounting model, the adoption of a principles-based accounting system under IFRS, and the impact of SOX as it moves closer to a state of equilibrium. • I also suggest that further research may be useful for better understanding the role of EQ in the debt markets, fraudulent financial reporting, policy making, and transactions management.

19. References • Dechow, P., Ge, W., Schrand, C., 2010. Understanding earnings quality: a review of the proxies, their determinants and their consequences. Journal of Accounting and Economics. • Jones, J., 1991. Earnings management during import relief investigations. Journal of Accounting Research 29, 193–228.